Disabled Medicare beneficiaries are among the costliest to the Medicare program. In addition to their substantial healthcare needs, disabled older adults often require long-term care (LTC). LTC provides help with daily activities (e.g., bathing, eating) that can be delivered in a facility like a nursing home (facility LTC) or in a community-setting like one's own home (community LTC). Though Medicare does not pay for LTC, facility LTC providers have increased access to healthcare providers and may have financial incentives to increase their residents' Medicare service use, particularly acute and post-acute care. However, it is unknown whether these incentives result in higher Medicare expenditures. The proposed research will determine whether facility LTC use increases Medicare expenditures and healthcare use for disabled older adults relative to community LTC. I will extend prior research by addressing the confounding often present due to high rates of comorbidities and other unobserved factors among facility LTC users. I will first seek to identify the independent effect of facility LTC on Medicare expenditures for acute care, physician services, and post-acute care. I will then seek to determine the effect of facility LTC on hospital related health services utilization, including emergency department (ED) visits, hospital observation days, and hospital readmissions. Methods: First, logistic regression will identify a clinically-similar group of community residents among the older adults in the Medicare Current Beneficiary Survey from 2000-2009. Incident users of facility LTC will comprise the treatment group and clinically similar community residents will comprise the control group. Second, instrumental variables will pseudo-randomize these groups to facility LTC to further adjust for any differences in co-morbidity and acuity between the groups. Number of adult children, local supply of LTC facilities, and state spending on home-delivered meals will act as instruments. Third, seven second stage models will be constructed, most using survival-adjusted methods. Survival-adjusted models are the most meaningful measure of cost in a population with high rates of death. There will be a model for each dependent variable: acute care Medicare expenditures, physician/practitioner services Medicare expenditures, post-acute care Medicare expenditures, number of ED visits, number of hospital observation days, and likelihood of 30- day and 60-day hospital readmission. Contribution and Significance: The proposed research seeks to determine the effect of a policy-modifiable factor, LTC setting, on the healthcare expenditures for high-cost Medicare beneficiaries. This will offer policymakers evidence to incorporate into payment policies as well as healthcare interventions targeted at high-cost beneficiaries. Meanwhile, the proposed research will apply novel econometric methods (including survival-adjusted methodologies) to LTC research. This will offer other researchers interested in the healthcare costs of the disabled an example of how to apply these exciting new methods to important policy questions.